This video demonstrates some techniques UW and Adobe are working on for deep analysis, object recognition, and embedding of metadata within video. It looks really interesting, and although at the moment it all looks a bit academic and probably requires a ton of processing time, but remember, that’s how Photosynth started and now it’s totally usable and has a growing community.
[via Reddit]